Why I've moved to Claude (and why your Coast business might too)

Most Coast business owners I talk to are on ChatGPT. That is fair. It was the first one most of us tried, it has had the biggest brand presence for two years, and the muscle memory is real.

Over the last year, a lot of professionals I respect have moved across to Claude as their primary tool. Some have cancelled ChatGPT outright. Others keep both, running Claude as the main platform and using ChatGPT for specific jobs. The shift was loud enough at the panels I sat on in West Palm Beach earlier this year that several speakers compared notes afterwards and realised we had all quietly moved.

This is the article I wish someone had handed me before I switched. It explains what Claude is, why the shift is happening, and how to move across in an afternoon without losing what you have built up in ChatGPT.

What Claude is

Claude is the AI assistant made by Anthropic, a company founded in 2021 by a group of former OpenAI researchers with a safety-first research mandate. When people say "I'm using Claude" they could mean a few different things.

claude.ai is the chat app on web, iPhone, Android and desktop. The closest equivalent to ChatGPT. Free tier for casual use, $20-a-month Pro plan for the full model range and the features below.

Claude Projects are organised workspaces where you load reference documents and set custom instructions for a particular kind of work. Each Project holds roughly 500 pages of documents, and every conversation inside the Project draws on them. For a Coast business this is a quietly powerful feature. A "Customer enquiries" Project loaded with your service list, pricing guide and tone notes. A "Bookkeeping" Project loaded with your chart of accounts and last year's BAS. The model stops being generic and starts being yours.

Extended thinking is a mode where Claude reasons through a problem step by step and lets you read along. For anything multi-step (a complex client question, a contract clause, a planning problem), it is the difference between an answer that feels confident and an answer you can trust.

Why a lot of professionals are switching

In the last twelve months Anthropic's share of the enterprise AI market has moved from around 12% to around 40%. That is not the fashion crowd. It is law firms, accounting firms, banks and government agencies choosing where to put their AI spend.

The reasons people give me are pretty consistent.

Writing quality. Claude's output reads more like a thoughtful colleague than a confident intern. For client-facing drafting, that matters.

Instruction following. When you tell Claude not to use a word, or to stick to a structure, or to keep an answer under a length, it tends to listen.

Reasoning on hard problems. Claude is currently rated very highly on complex, multi-step problems. If your work involves anything where the answer has to be defensible, this shows up.

Honesty about uncertainty. Claude is more willing to say "I am not sure" or "I do not have enough information to answer that well". For a regulated context, that is a feature, not a bug.

None of these alone is decisive. Together, for the kind of work most professional services on the Coast do, they add up.

How to move across without losing your ChatGPT context

The most common reason people give for staying with ChatGPT is the months of context they have built up in it. You do not have to lose it.

Anthropic built a memory import tool directly into claude.ai. The process takes ten minutes:

1. Open claude.ai > Settings > Capabilities.
2. Find "Import memory from other AI providers" and follow the steps.
3. Claude generates a short prompt for you to copy into ChatGPT.
4. ChatGPT returns a structured summary of what it remembers about you (preferences, working style, key facts).
5. Paste that back into Claude. Your working context transfers across.

Review what comes through before you accept it. AI memory is a curated list of preferences, not a transcript of every conversation. Top it up manually with anything specific you want Claude to carry forward.

The one privacy setting to flip before you load anything

If you are going to put any client information, commercial detail or business data into Claude, do this before you do anything else.

Go to Settings > Privacy in claude.ai. Find the toggle labelled "Help improve Claude". Switch it off.

By default, Anthropic may use your conversations to train and improve its models. For casual personal use, some people are fine with that. For a Coast business handling client details, customer enquiries or financials, you want this off. It takes thirty seconds. It applies to your whole account, not each conversation, so you only have to do it once.

This applies to individual and Pro plans. If you go on to use a Claude Team or Enterprise plan later, the data terms are different by default and an admin should review them separately.

A governance note worth saying out loud

I would not recommend anyone build their whole business around a single AI provider, including Claude.

Vendor concentration in AI is a real strategic risk. Keep an account on at least two platforms. Understand what each does well. Run your important workflows in a way that does not melt if one provider has a bad week. I have written about this in more depth in The model wars. The short version is: pick your favourite, keep a backup, and stay on monthly billing.

A first afternoon with Claude

If you want to try it properly, here is what an honest first afternoon looks like.

1. Sign up at claude.ai. Pro plan, monthly, $20.
2. Settings > Privacy. Switch off "Help improve Claude".
3. Settings > Capabilities. Import your memory from ChatGPT.
4. Create one Project for the part of your business you are in most often. Upload three to five reference documents. Set a one-paragraph custom instruction about the business and how you want it to write.
5. Run the next real piece of work you would have run in ChatGPT through Claude instead. Compare.

Within a week of regular use, you will have a clear answer about whether the switch makes sense for you. There is no wrong outcome. The point is that you have actually tested it, with your own work, on your own data, and made an informed call.

If you would like a hand setting Claude up properly for your business (the Projects, the custom instructions, the privacy settings, the workflow design), our AI workspace setup is the engagement for that. The Find your AI quick wins email is the right first move.